Factors Affecting Seed Germination of the Mauna Kea Silversword in Hawai'i1

LAWRENCE R. WALKER 2 AND ELIZABETH ANN POWELL 3

ABSTRACT: The Mauna Kea silversword, Argyroxiphium sandwicense DC, is endemic to the slopes of Mauna Kea on the island of Hawai'i. Once abundant, it is now reduced to a total of less than 500 individuals. We examined germi nation of silversword seeds subjected to various experimental and field condi tions. Under experimental conditions, germination was optimal in moist, shady environments. Removal of the pericarp greatly enhanced germination, but cold and heat pretreatments did not alter germination. Germination offield-collected seeds was highest for seeds < 2 yr old and for seeds collected from flower stalks. The ability to germinate was much lower for seeds collected from on or under the soil surface. In mesic environments, grasses competed with silversword seedlings. We suggest that seed germination and early seedling establishment are major obstacles to reestablishment of the Mauna Kea silversword.

THE MAUNA KEA SILVERSWORD (MKSS; Ar rocephalum A. Gray), a threatened species gyroxiphium sandwicense DC) is a federally endemic to Haleakala Crater on the island of listed endangered plant and is endemic to the Maui. The Haleakala and Mauna Kea sil alpine slopes of Mauna Kea on the island of verswords are variously considered subspecies Hawai'i. The current population consists of (Carr 1985, Wagner et al. 1990) or separate about 40 naturally occurring individuals in species (Degener, pers. comm.; Powell 1992) one population and about 400 individuals based on morphological characteristics. Ge that are the result of outplanting at three lo netic analyses ofvariation between and within cations by the Hawai'i State Division of the Haleakala and Mauna Kea populations Forestry. The largest population of outplants are currently in progress (R. Robichaux, pers. is located on the eastern slope of Mauna comm.). This study considers the MKSS as a Kea between 2800 and 2900 m elevation at distinct species (Powell 1992). Waipahoehoe Gulch. Two smaller outplanted Recovery of the MKSS to population populations are located on the western slope levels that will allow it to be removed from of Mauna Kea at Pu'u Nanaha (2770 m) and the endangered species list is mandated by along the Skyline jeep trail (2970 m). The the Endangered Species Act (Rohlf 1991). MKSS is a giant rosette plant of the family Two approaches are being attempted to Asteraceae, and individuals are either mono reestablish the MKSS. The Hawai'i State carpic or polycarpic (have one or several ro Division of Forestry has conducted an ex settes [Powell 1992]). Showy flower stalks are tensive program of outplanting shadehouse produced that reach 2-3 m in height. The grown seedlings on Mauna Kea (l974-pres MKSS is less well known than its relative, the ent [powell 1992; S. Bergfield, pers. comm.]). Haleakala silversword (Argyroxiphium mac- However, outplanting is costly and labor in tensive and survival rates of outplants have been low (14-42% survival in 1987 [E.A.P., 1 This project was partially supported by grants from unpubl. data]). In addition, shadehouse- or The Nature Conservancy and the V.S. Fish and Wildlife greenhouse-grown plants may be selected for Service. Manuscript accepted 22 September 1994. characteristics not common in the natural 2 Department of Biological Sciences, V niversity of Nevada, Las Vegas, 4505 Maryland Parkway, Las populations (Powell 1992). The second ap Vegas, Nevada 89154-4004. proach is to examine the feasibility of sowing 3P.O. Box 61872, Boulder City, Nevada 89006-1872. seeds directly at potential regeneration sites 205 206 PACIFIC SCIENCE, Volume 49, July 1995 to avoid the problems associated with out surface ofwet vermiculite (288 achenes total). planting. In both approaches, artificial cross Within each shade treatment, 24 pots were pollination of flowers in the field is necessary watered daily (unless it rained) and 24 were to obtain adequate numbers of filled seeds watered weekly. Seed germination (> 1 mm because the MKSS is self-incompatible and elongation of the radicle) was monitored seed set is pollinator-limited (Powell 1992). daily for 33 days (7 days after last germina In this paper we report on experiments tion occurred). The rate of germination was that examined the germination of MKSS measured as the number of days to reach seeds from achenes subjected to greenhouse 50% of final germination (GT50)' Statistical and field conditions. Specifically, we looked differences among treatments were deter at the effects of various shade and water re mined at the 0.05 level using chi-square anal gimes, pericarp removal, and cold and heat ysis of contingency tables on the number of treatments on germination of MKSS seeds in seeds that germinated. the greenhouse. We also compared germina PERICARP REMOVAL EXPERIMENT. Seventy tion of seeds that remained on flower stalks, two achenes with filled seeds from the same fell to the ground, or were buried in the soil. September 1987 collection were removed from Finally, we examined the numbers of seed storage at 5°C and soaked in tap water for 3 lings established from MKSS achenes sown hr before being subjected to one of three dif in different microhabitats in the field. The ferent treatments (n = 24 achenes per treat results of these experiments will help deter ment): total, partial (one-fourth to one-fifth mine if sowing MKSS achenes in the field is a removed), or no removal of the pericarp (the viable alternative to outplanting and what hardened fruit wall). The achenes were placed the optimal conditions are for germination. on the surface of wet vermiculite in a green house at HAVO and misted four times daily to maintain a moist surface. Seed germination was monitored for 33 days. MATERIALS AND METHODS COLD AND HEAT EXPERIMENTS. Achenes Greenhouse Experiments with filled seeds from the same source as In July 1988 we conducted three experi previously described were subjected to five different temperature regimes (room tem ments designed to determine factors affecting germination of MKSS seeds. The experi perature [about +20], -15, + 30, +55, and 75°C) for various lengths of time (30 min ments were conducted at Hawai'i Volcanoes + National Park (HAVO) on the island of to 48 hr; total of 11-24 achenes per treat Hawai'i (1200 m elevation). Achenes (fruits) ment, 16 treatments, 339 achenes total). After were collected in September 1987 from 20 the heat treatments, achenes were soaked in flowering silverswords from the Waipahoe tap water for 3 hr and the pericarp was par hoe and Skyline populations on Mauna Kea tially removed. Achenes were then sown on a and stored at 5°C. Contributions from indi surface of wet vermiculite in pots at HAVO vidual plants varied between 0.5 and 10% of and misted four times daily. Seed germina the total seed collection. Only achenes with tion was recorded daily for 33 days; treatment filled seeds (containing intact embryos) were differences were evaluated with chi-square used for the three experiments. analysis.

SHADE AND WATER EXPERIMENT. Forty eight pots were placed in each of six shade Field Experiments treatments (0, 30, 55, 63, 80, and 96% reduc SEED GERMINATION: COLLECTIONS. Germi tion of ambient light by optically neutral nation of seeds from stalks was determined shade cloth) on open-air benches at the from achenes collected from a total of 13 greenhouse complex at HAVO. One achene flower stalks in three populations (Waipa (with filled seed) was sown in each pot on a hoehoe, Pu'u Nanaha, and Skyline) in July

EgA !If' stza::zuaz:awumW#&&P k'Ai!P¥## ,tM:w:et·#!4ffiM@@; gMJI 'lim,,' @4&!.e 4i44lJ!i;4Z& Seed Germination of the Mauna Kea Silversword-WALKER AND POWELL 207

1988. The stalks were either 1 (n = 6 stalks), SEED GERMINATION: EXPERIMENT. To deter 2 (n = 4), or 3 (n = 3) yr old. In July 1989, mine the effect of burial on seed germination, achenes were collected from an additional six 10 achenes with filled seeds from the same 1-yr-old flower stalks at Waipahoehoe (total September 1987 collection were placed in of 19 stalks). We did not collect from plants nylon mesh bags (10 by 7 cm; mesh size of that flowered in multiple years or that were in 20 threads per centimeter). The bags were close proximity to other flowering plants so closed with wire ties and in July 1988 placed age of achenes could be determined. When either on the surface of the soil at 1.5, 2.5, possible, five heads (capitula) with achenes or 3.5 m or under 1 cm of soil at 1.0, 2.0, or attached were collected from each flower 3.0 m from 12 of the 13 1988 flower stalks stalk, and all achenes were sorted to deter described above (total of 72 bags and 720 mine percentage of filled seeds. Germination seeds). Bags at ~2 m distance from the plant of seeds was then tested. Achenes were first were retrieved after 1 yr; bags < 2 m from the soaked in water for 3 hr, the pericarp was plant were collected after 2 yr. Achenes were then partially removed, and the achenes were pooled by treatment (surface or buried) and placed on moist vermiculite in greenhouse date of collection for chi-square analysis, pots and misted daily. Germination was re and seed germination was determined as corded for 33 days. previously described. Germination of seeds and percentage of achenes with filled seeds from the ground FIELD SOWING EXPERIMENT. This experi surface were determined from achenes col ment was designed to determine the effect of lected within a 1-m radius of (and in a organic soil versus cinder and the presence or northerly direction from) the same 19 flower absence of shade and grasses on MKSS seed stalks. ling establishment. Two thousand achenes Germination of seeds found in the soil was with filled seeds were obtained from the Sep also determined. Two bags of soil, each 5 by tember 1987 collection on Mauna Kea (1000 10 by 1 cm deep, were collected around each from plant no. 23 at Skyline; 1000 from of the same 19 flower stalks (13 in 1988, 6 in mixed sources). On 21 July 1988, 50 achenes 1989) at each of three locations: directly with filled seeds were sown on the surface under the old flower stalks, at 50 cm, and at of the soil in 10 by 10 em plots located in 100 cm distance from the flower stalks in a 10 areas (blocks) with four treatments per southerly direction. These soil samples in block (a total of 40 plots and 2000 seeds). cluded any achenes that were on the soil sur The treatments were under mamane (So face as well as those achenes buried to 1 cm phora chrysophila St. John) trees with 100% depth. The soil was sifted for achenes, and grass cover; under mamane with removal of percentage achenes with filled seeds and per all grass from a 25 by 25 cm area, leaving ex centage seed germination were determined. posed organic soil; in adjacent open habitat These data were also used to estimate density on cinder without grass; and on adjacent of achenes on and in the soil. The surface open cinder habitat with 25-75% grass cover. 2 area of the two bags (100 cm ) was used to The blocks were located at least 30 m apart extrapolate achene density in each of three in a 500 by 100 m area within the Waipahoe concentric zones around each flower stalk. hoe Gulch silversword exclosure on Mauna The sampled radius and area of each of the Kea. Each treatment within a block was lo 2 three zones was 0-25 cm (1963 cm ), 25-75 cated randomly within the appropriate habi 2 cm (15,708 cm ), and 75-125 cm (31,416 tat along a 10- to 15-m transect and radiating 2 cm ), respectively. Data from the 19 stalks out in a northeasterly direction from large were pooled by stalk age before germination mamane trees. was compared by stalk age and habitat (stalk Achenes with filled seeds from the same or soil) with chi-square analysis; density of sources and year were also sown on wet ver achenes was compared with a Kruskal-Wallis miculite in a greenhouse in Volcano, Hawai'i test. (1200 m elevation), and misted daily for 208 PACIFIC SCIENCE, Volume 49, July 1995

33 days to test viability of the field-sown TABLE 2 achenes. Seedling establishment in the field GERMINATION OF SILVERSWORD SEEDS UNDER VARIOUS was recorded in July 1989 (1 yr) and Sep TEMPERATURE REGIMES tember 1990 (2 yr) as a percentage of achenes with filled seeds sown and compared with chi TREATMENT GERMINATION square analysis. DESCRIPTION (temperature [oq, NO. time") nb % RATE (GTso) RESULTS 1. ambient, 0 hr 23 87 5 2. -15°,24 hr 24 92 3 Greenhouse Experiments 3. -15°,48 hr 24 92 4 4. +30°,30 min 24 92 5 SHADE AND WATER. Germination of seeds 5. +30°,60 min 24 92 5 watered weekly differed significantly (chi 6. +30°,6 hr 22 77 5 square = 16.9; P = 0.002) among shade treat 7. +30°,24 hr 24 92 7 ments and was highest at intermediate and 8. +30°,48 hr 24 92 6 high shade levels (Table 1). Germination of 9. +55°,30 min 23 96 5 10. +55°,60 min 23 78 4 seeds watered daily was not significantly 11. +55°,6 hr 21 95 4 different (chi square = 4.7; P = 0.40) among 12. + 55°,24 hr 24 87 4 shade treatments. Watering regime made less 13. + 55°,48 hr 24 96 5 difference at higher shade levels, presumably 14. +75°,30 min 12 92 5 15. + 75°,6 hr 12 92 5 because shade reduced water loss. Germina 16. + 75°,48 hr 11 91 5 tion rates (GT50) were fastest at higher levels of shade (Table 1). "Time in treatment after removal from 5'C and before sowing. PERICARP REMOVAL. Removal of the total b n ="11-24 achenes with filled seeds. pericarp resulted in the highest (87%) and fastest (GT50 = 7 days) germination, fol lowed by partial (71%, 8 days), and no re COLD AND HEAT. Germination percentages moval (42%, 18 days). The pericarp clearly did not differ from the control at ambient inhibited germination of the MKSS seeds. temperature (chi square = 10.7; P = 0.70) following a wide range of temperature pre treatments (Table 2), suggesting that MKSS TABLE 1 seeds are tolerant of freezing or heating for short periods of time. GERMINATION OF SILVERSWORD SEEDS UNDER SIX SHADE AND Two WATERING TREATMENTS Field Experiments GERMINATION SEED GERMINATION: COLLECTIONS. The per % centage of achenes with filled seeds (Table 3) % declined with stalk (and therefore achene) SHADE DAILY" WEEKLY DAILY WEEKLY age (chi square = 97.1; P < 0.0001) and was 0 46 12 21 21 largest for achenes collected directly from 30 62 46 19 19 stalks and smallest for achenes collected from 55 58 nd 16 nd in the soil (all three distances from stalk com 63 75 58 13 14 80 62 67 17 14 bined; chi square = 89.0; P < 0.0001). Per 96 54 54 14 13 centage germination of seeds that remained on the stalks was > 70% after 1 or 2 yr. Only NOTE: Twenty-four achenes with filled seeds were sown per one filled seed was found on the 3-yr-old treatment combination. nd, no data available. "Watering regime. stalks that were sampled, and that seed did bTime in days to reach 50% of final germination. not germinate. Seed survival is therefore pos-

• i Ii@A W t# 41 ,,'i' ..;;$.'$ AWe ,a Seed Germination of the Mauna Kea Silversword-WALKER AND POWELL 209

TABLE 3

GERMINATION OF SILVERSWORD SEEDS COLLECfED FROM STALKS, FROM THE GROUND SURFACE, OR FROM THE Top I CM OF SOIL

% FILLED SEEDS % GERMINATION OF FILLED SEEDS

STALK NO. ON ON SOIL ON ON SOIL AGE (yr)a STALKS STALKS SURFACE IN SOIL STALKS SURFACE IN SOIL

12 2.9 2.5 1.9 70.5 60.0 22.2 (2,667) (798) (465) 2 4 1.5 0.8 0.0 75.0 100.0 NA (533) (247) (101) 3 3 0.4 0.0 0.0 0.0 NA NA (238) (85) (71)

NOTE: Seed age equals stalk age. Numbers of seeds collected are in parentheses. NA, not applicable. "From July 1988 collection (stalk ages 2 and 3 yr) and both July 1988 and July 1989 collections (stalk age 1 yr). sible on the stalks or soil surface for 2 yr but 19.7; P = 0.04) for achenes buried in the field probably not for 3 yr. Burial reduced seed in mesh bags (38%) than for achenes placed survival to a maximum of I yr. in mesh bags on the soil surface (28%). Re Density of achenes with or without filled trieval of filled seeds after 2 yr was minimal seeds on or in the soil declined rapidly with for achenes that were buried (two seeds distance from flower stalks (Kruskal-Wallis [1.7%]) and for achenes on the soil surface H = 24; P < 0.0001), but did not decline con (three seeds [2.5%]). Germination of surviv sistently with stalk age (Kruskal-Wallis H = ing seeds was not significantly higher (chi 0.8; P = 0.68 [Table 4]). However, achenes square = 17.6; P = 0.09) for buried achenes that had been in the soil for 3 yr were proba in year 1 (buried: 83%; surface: 65%). In year bly not viable (Table 3). 2, one of the filled seeds germinated from the buried achenes and two from the surface SEED GERMINATION: EXPERIMENT. After 1 achenes. All but 5% of the achenes were re yr, the percentage of achenes that still con covered, but most had been damaged by tained filled seeds was greater (chi square = fungi, broken, dried out, or contained seeds that germinated and then died. TABLE 4 FIELD SOWING. One year after sowing, DENSITY OF SILVERSWORD ACHENES AT THREE DISTANCES seedling establishment of silverswords was FROM FLOWER STALKS highest (chi square = 27.0; P = 0.003) under mamane with grass removal (4.2 ± 1.7% 2 ACHENE DENSITY (no./em ) [mean ± SEM] of the viable seeds sown; 12

STALK NO. seedlings total in five plots) followed by AGE (yr)' STALKS 0-25 ema 25-75 em 75-125 em open cinder with grass (3.6 ± 2.5%; 10 seed lings in two plots). Only one seedling was 1 12 0.35 ± 0.15 0.04 ± 0.01 0.03 ± 0.01 found under mamane with grass, and no sil 2 4 0.13 ± 0.01 0.04 ± 0.01 0.08 ± 0.07 verswords established in the open habitat 3 3 0.18 ± 0.07 0.03 ± 0.01 0.03 ± 0.01 without grass. No seedlings survived for 2 yr. NOTE: Densities were calculated using number of achenes with or without filled seeds from the soil surface and top I em of soil divided by the area of each of three concentric bands around the stalks. DISCUSSION a Range of radial distances from flower stalk. 'From July 1988 collection (stalk ages 2 and 3 yr) and both Germination may be a major factor limit July 1988 and July 1989 collections (stalk age I yr). ing reestablishment of the Mauna Kea sil- 210 PACIFIC SCIENCE, Volume 49, July 1995

versword. Under experimental conditions, the seeds were in the soil than when they germination was optimal in moist, shady remained on the flower stalks. Wetting of environments. However, establishment of achenes in the soil may make the seeds more MKSS seedlings under the shade of large susceptible to extreme temperatures, desicca trees was apparently prevented by dense cover tion, or attack by fungi. Primary causes of of grasses. Competition between grasses and seed mortality on flower stalks appeared to seedlings of long-lived alpine rosette plants be seed herbivores (notably tephritic fly lar has been found in other studies (Smith 1984, vae that parasitized the developing embryos Augspurger 1985). [Powell 1992]). The pericarp (fruit wall) of the MKSS Dispersal of the MKSS achenes appeared clearly inhibits germination, probably due to to be primarily limited to 1-2 m from the chemical inhibitors that Siegel et al. (1970) parent plant. Long-distance dispersal by detected in extracts from both achenes and water or wind is possible but has not been inflorescences of the related HaleakaHi silver documented (beyond one example of appar sword. Germination was highest in our study ent water dispersal downslope a distance of when the entire pericarp was removed. Partial 8 m [L.R.W., pers. obs.]). We found no evi removal of the pericarp resulted in inter dence of animal dispersal. In the absence of mediate germination. When the pericarp was active dispersal mechanisms, germination of only punctured (Siegel et al. 1970), germina the few filled seeds (0.5-5% of the total tion was not enhanced. Germination of the [Powell 1992]) may be inhibited by leachates MKSS probably occurs following natural from the pericarps of thousands of achenes weathering of the inhibitory pericarp. This with unfilled seeds that also fall near the weathering may occur most rapidly in mesic parent plant (see Fenner 1985). environments. Sowing MKSS achenes in optimal micro Dry MKSS achenes tolerated extreme tem sites away from the potentially inhibitory ef peratures for short durations in our study. fect of parent plants and other achenes may Wide ranges in soil surface temperatures are improve germination in the field and prove to typical of Hawaiian alpine areas, with freez be a less costly and more logistically feasible ing temperatures common at night and day option than outplanting. Because the seeds time maxima exceeding 40°C (Kobayashi have no dormancy requirement (typical of 1973). Yet Siegel et al. (1970) found that monocarpic plants [Young and Augspurger germination of the Haleakala silversword 1991]), achenes may be sown immediately was heat sensitive, with best germination (of after collection. Seed viability declined rap imbibed seeds) at alternating temperatures of idly in the soil, so collections should be made 8 and 15°C, and no germination in any re directly from the stalks. Further experiments gime (continued for 3 weeks) involving 8 hr are under way to evaluate the effects of ele or more at temperatures of -15 or +35°C. vation and other microsites on germination. Our study did not have a comparable 3-week Elevation, for example, can affect micro temperature regime, and we only used tem climate and the morphology, physiology, and perature as a pretreatment, subsequently reproduction of alpine plants (Baruch 1979, placing seeds in a moist environment at am Smith 1980). Powell (1992) noted higher bient temperature to germinate. However, production of achenes with filled seeds but Siegel et al. (1970) found that germination lower seedling establishment for plants at was reduced after a pretreatment for dry the Skyline exclosure than for plants at the achenes of just 15 min at 60°C. Our results lower-elevation Waipahoehoe Gulch. showed no effect on germination even after a pretreatment of 48 hr at 60°C; Removal of the pericarp after the heat treatments may ACKNOWLEDGMENTS have enhanced germination of heat-treated seeds. We thank R. Robichaux and L. Stemmer Seed viability declined more rapidly when mann for stimulating discussions.

ii'4 ii;; ou :au Seed Germination of the Mauna Kea Silversword-WALKER AND POWELL 211

LITERATURE CITED ROHLF, D. J. 1991. Six biological reasons why the endangered species act doesn't AUGSPURGER, C. K. 1985. Demography and work-and what to do about it. Conserv. life history variation of Puya dasliriodes, a BioI. 5: 273-282. long-lived rosette in tropical subalpine SIEGEL, S. M., P. CARROLL, C. CORN, and T. bogs. Oikos 45: 341-352.. ... SPEITEL. 1970. Experimental studies on the BARUCH, Z. 1979. Elevation dIfferentIatIOn In Hawaiian silverswords (Argyroxiphium Espeletia schultzii (Compositae), a giant spp.): Some preliminary notes on germi rosette plant of the Venezuelan paramos. nation. Bot. Gaz. 131 :277-280. Ecology 60: 85-98. SMITH, A. P. 1980. The paradox of plant CARR, G. D. 1985. Monograph of the Ha height in an Andean giant rosette species. waiian Madiinae (Asteraceae): Argyroxi J. Ecol. 68: 63-74. phium, Dubautia, and Wilkesia. A11ertonia ---. 1984. Postdispersal parent-offspring 4: 1-123. conflict in plants: Antecedent and hypoth FENNER, M. 1985. Seed ecology. Chapman & esis from the Andes. Am. Nat. 123: 354 Hall, New York. 370. KOBAYASffi, H. K. 1973. Ecology of the sil WAGNER, W. L., D. R. HERBST, and S. H. versword Argyroxiphium sandwicense DC SOHMER. 1990. Manual of the flowering (Compositae), Haleakala Crater, Hawaii. plants of Hawai'i, 2 vols. Bish~p M,useum M.S. thesis, University of Hawai'i at Special Publication 83. Umverslty of Manoa, Honolulu. Hawai'i Press, Honolulu. POWELL, E. A. 1992. Life history, reproduc YOUNG, T. P., and C. K. AUGSPURGER. 1991. tive biology, and conservation of the Ecology and evolution of long-lived Mauna Kea silversword, Argyroxiphium semelparous plants. Trends Ecol. EvoI. 6: sandwicense DC (Asteraceae), an endan 285-289. gered plant of Hawaii. Ph.D. diss., Uni versity of Hawai'i at Manoa, Honolulu.

..~,